Sharing Calibration Settings: Results Compared
By Chris Heinonen on
Often when I write a review of a display or projector, someone asks me for my specific display settings. I typically mention which color temperature or picture mode works best, but many readers want specifics. How did I adjust the white balance controls? Did I make changes to the color management system? What about a 10-point white balance control?
I’ve always refused to give these specifics. Just like any product, there will be variations between two samples of the same display. I have no way to know if my settings will help or hurt a display, and most readers have no way to know if this will help them or not. Sharing calibration settings, and if it helps or hurts, is one area I’ve wanted to look into for a long time.
I just purchased a Panasonic VT60 plasma display, as did Ray Coronado, a calibrator in the Southern California area. Ray wanted to try to measure the results of using publicly available calibration settings on the VT60 with our individual units, and I wanted to know as well. We decided to take settings from two of the most used sources: D-Nice, a professional calibrator who has posted his data for years, and CNet, where they also post their settings.
For these measurements we made sure to break-in our displays for 100 hours using the slides that D-Nice provides. For all measurements we use 2% windows except for CNet as they use 18% windows for their calibrations and measurements.
I used an i1Pro and C6 while Ray used an i1Pro2 for readings. We also both measured against the default THX modes in our displays for reference points. We chose a gamma target of 2.2 as that is what CNet and THX both use, and what it seems that D-Nice also uses. The exception is for the THX Bright Room setting that seems to target a gamma of 1.9.
Here are my results first for all three copied settings and the two preset numbers.
|D-Nice Day||D-Nice Night||CNet||THX Cinema||THX BrightRoom|
|Light Output||43.3 fL||33.6 fL||38.6 fL||32.2 fL||53.9 fL|
|Color Checker dE2000||1.85||2.71||1.31||1.75||1.88|
What stands out here is that the best overall settings are possibly from CNet. The gamma is a little off, but the overall errors are lower than any other setting. It also uses the Cinema mode as a preset instead of using the Professional1 or Professional2 modes that D-Nice does. Once we look at these in more detail later you’ll see why I would never actually use the settings from CNet on my display.
Looking at the THX preset versus the D-Nice settings there is a lower Grayscale dE2000 from D-Nice, but higher color errors. Those are worse in the nighttime setting, which is more directly comparable to the THX Cinema setting than the Day mode is. The errors in THX Bright Room mode are overall very low. I think the gamma is too low, but the preset is very good overall.
Now we can look at the results that Ray has from his VT60.
|D-Nice Day||D-Nice Night||CNet||THX Cinema||THX BrightRoom|
|Light Output||40.7 fL||31.6 fL||36.6 fL||32.0 fL||50.6 fL|
|Color Checker dE2000||2.41||2.67||1.51||1.58||1.81|
Ray has higher error levels than I do with the D-Nice Day settings; though his night results are very close to mine. He has a better gamma with the CNet numbers, and we both have similar results from THX.
Two things stand out when I compare these. The first is that the light output from the settings is off by 5-10% between the two sets. Even with two sets of the same model, bought the same day from the same store, the light output can vary by this much with the same setting. We also see that copying the calibration settings from D-Nice gives us decent results, but not ones that are any better than the THX Cinema mode. They also vary a lot between the two displays. Now we want to look at these in more depth, to see some of the other differences between the two.
One note here: Ray used his i1Pro2 to measure everything so the 5% and 10% readings are only so reliable. It does not affect the overall results, as every preset is subject to the same issue, but if you notice a gamma dip at 5% or 10% the reason is more likely the instrument being used and not the display.
The best RGB Balance of any setting is the CNet settings on my display. The THX Cinema is good on both but leans to a bit too much red. On Ray’s set the CNet settings lean towards too little red more than mine do. On both sets the D-Nice settings have too little red in the RGB Balance. So perhaps the set, or sets, that D-Nice used to get these settings has more red in it by default than our sets do. So by copying his settings we are introducing even more red into the grayscale.
CNet did not use the 10-point gamma control to adjust their settings, so their gamma uses the defaults. It is a bit low for both of us, with dips at 5% and 90% for both of us. Overall the D-Nice gamma settings track quite well to the 2.2 target that he is after. The THX Bright Room setting tracks very close to 1.9 for both of us, with the same bump and dip on both sets. Both sets see a dip at 5% with THX Cinema mode, but otherwise good tracking for a target of 2.2. Once again I do not worry about the 5% readings here for Ray as that’s an i1Pro2 issue. You can see that the THX Bright Room mode tracks better because it has a lower gamma. This lets the i1Pro2 read the 5% sample better resulting in a flatter gamma.
When we look at the dE2000 error levels for the grayscale overall, we see they are almost invisible errors. A couple of points might pop up above the magic level of 3.0 but mostly they are well below it. The THX settings do just as well as copied calibrated settings here. For Ray, the THX error levels are better overall than the ones he achieved using the copied calibration settings. If you are hoping to improve the grayscale quality by copying settings instead of using the display defaults, you really aren’t going to get any benefit.
As mentioned earlier, the CNet grayscale quality is very high. It uses the Cinema mode instead of the Professional1 or Professional2 modes that D-Nice uses. Entering their settings they left the color settings alone except for an adjustment to the magenta. It seems that there is something different with their sample compared to ours, as you will see in the saturations charts below. Sample variations like this is a reason that many people do not like to share their settings. It is impossible to know if this flaw is specific to your set, or to the product.
Overall for the saturations we found that the THX modes were more accurate than copying settings, with the THX Bright Room mode being really accurate. D-Nice’s settings still gave us a result where more errors were below a dE2000 of 3.0, but higher than the THX Mode. The CNet settings are good except they have a huge flaw in magenta. There is no reason to use copied settings over the THX defaults here.
We also tested with the large Color Checker chart included in CalMAN 5.2. This tests a large number of samples that are not primary or secondary colors, but instead colors you find in the real world. Sky blues, natural greens, skin tones, and more. This is a more accurate test of the color management system in the display as many can get primary and secondary colors right, but fail when it comes time to do colors between those points.
With these results it looks like the CNet numbers are the best, except for Magenta which causes a few large errors. These errors are well past the visible error limit, so they will show up. The difference between a dE2000 of 1.0 and 2.5 is typically not visible, while the one between 2.5 and 7.0 will be visible. As low as the overall errors are for the CNet settings, those few colors having large errors makes it the worst choice to me.
Looking at the D-Nice settings they are low overall errors, except they are still worse than the THX modes. The work of inputting these settings isn’t adding any benefit to the display and is adding a level of randomness that might not exist with the THX mode.
One interesting note is that on Ray’s THX Cinema mode, one skin tone kept registering an error level over 19.0 for him. He measured it five times and still got the same result. He didn’t notice the error in real world watching, so we tend to think it is a software glitch over the display being so far incorrect here. However we are not certain of this. If this reading is wrong, than his overall Color Checker dE2000 error level would fall by 0.15-0.18.
As we have seen, copying calibration settings did not do anything to improve upon the THX modes that are inside the VT60. With sample variations, what settings might work for someone will not necessarily work for another person. Both Ray and I achieved slightly different results with the same settings and a TV set ordered from the same company on the same day. On a set made earlier in the year, perhaps those settings worked better but we do not know.
After doing these measurements I calibrated my VT60 in Professional1 mode for myself. Unlike the preset modes, I want a gamma of 2.4 so I have darker shadow detail when I watch without any ambient light. Otherwise I used the same targets that I did earlier.
As you can see the overall error levels are incredibly low. None of the issues seen when using copied settings are present and there are no visible flaws in this image. With a full 6-point CMS that offers Hue, Saturation and Color adjustments, as well as a 10-point white balance and gamma controls, the VT60 offers the controls you need to hit your reference targets. This is an improvement over the standard THX settings and how I will run my display at night when it is totally dark.
What do I take away from this? We only have a pair of displays available to us to test, so this is a very small sample. In our cases, copying settings does not improve upon the THX modes. For almost everyone those THX modes are accurate enough that they will be very happy to use them and never want anything more. If you want a more accurate image you can get one but only by calibrating that individual display. Just copying the adjustments someone else made to the same model will not make yours better.
Now when asked why I don’t share my calibration settings, I will simply point back to this article. Compared to the pre-calibrated settings from the factory mine are not likely to improve the set and will probably only make it worse. If you really want the most accurate set you can get, then you’ll need to have it individually calibrated as every display is unique.